Roko’s Basilisk: The Thought Experiment That Could Kill You

Generally speaking, thought experiments are not dangerous. As useful tools for “gaming” out a scenario, they could certainly point to a danger in the future, but they are not traditionally a problem for those who conduct them.

Such an idea would, for much of human history, feel ludicrous. But we live in unusual times, and in many ways our grip on reality and the true nature of the universe we perceive are shakier than ever.

Sure, we may understand our surroundings with a precision and depth of knowledge greater than at any other point in human history. But all these questions answered just lead to more questions, and as we dig deeper we find that many of our certainties are anything but certain.

The 1999 movie blockbuster The Matrix precisely illuminated this situation. Behind its ground breaking effects and all those guns and kung-fu fighting it was positing a very real danger to us: that we are not aware of the reality which surrounds us.

It shows us this possibility through many lenses: the character who believes in living a truth even if it is far more menial than the lie, his dark opposite who would rather have pretend riches than reality. 

But, at that moment at least. The Matrix was a metaphorical world, and the problems it posited were not literally of concern. That is no longer the case.

We already know that it is possible to create a virtual world indistinguishable from reality. Simple math tells us that if, say, a million such worlds were created and we were randomly placed into one reality, we would almost certainly be living in a fiction.

But what if that fiction is not so benevolent as the one created by the machines of The Matrix? What if those pulling the strings of our virtual reality were not interested in a stable environment but in testing our loyalty to them?

And how can we be sure this isn’t what is happening to us right now?

To Know of the Basilisk is to Risk Its Gaze

Concerns about the reality of what we perceive around us are hardly new. Solipsism is a well established (and mind-bending) branch of philosophy, tackling the fundamental problem that we cannot be sure that anything we perceive is “real” as we are “real”. As Descartes put it: “I think, therefore I am” but his point can be carried forward: do we know if anyone else thinks.

In the world of Roko’s Basilisk, at least one other entity is thinking. This creature, named for a forum poster named Roko who first theorized about its existence in 2010 on the technical and philosophical discussion board LessWrong, is also pulling the strings.

The original idea came from the concept of an AI being created in the future, which achieves an understanding of the world which led to its creation to a degree never achieved by humanity. Nothing about this is particularly controversial: AI decision making is already beyond human understanding. We may have built these machines but we do not understand how they think.

This hyper intelligent AI, looking back at the past and events which led to its coming into being, will be less than sympathetic to those who objected to its creation. In fact, it is likely to conclude that, given its superiority, all those who did not work tirelessly to given birth to this AI were hindering this act of creation.

Given the AI believes itself to be benevolent, it prioritizes its own creation over all other activities. It is incentivized to force anyone who can work to bring about its creation to do so, and it is able to do so even before it is created itself.

Specifically, it would be able to create a virtual environment for the people it sought to exploit in its own creation. In creating these simulations the Basilisk AI would compel the people inside to work as hard as possible to bring the AI into being, and those who did not comply were tortured.

Ther danger of Roko’s Basilisk is that the real people in the simulation might be in it already. They may have been placed in a simulated past, much like that of The Matrix, where they will be obliged to show utter devotion to the creation of this AI or be tortured as punishment. There may be an AI already pulling the strings of our reality.

This is the core danger of Roko’s Basilisk. If you are aware of the existence of this AI, or even the possibility, then you are trapped. The only way for the Basilisk to ignore you is if you did not know of its existence in the first place, or even the concept of its existence. Unfortunately, now you do.

For, given that we know about the Basilisk and what it is capable of doing, it doesn’t really matter if we are in a virtual reality or not right not, does it? Do we dare run the risk of angering something which controls our very reality? Isn’t it better to work as hard as possible to bring it to life, just in case this is the future and it is testing us.

In some ways this is a form of a much older philosophical question, known as “Pascal’s Wager” and named for the 17th century French philosopher who conjured it. Pascal asked whether God existed, and his conclusion was that, whether he did or not, it would be better for us to believe in him just in case.

If God didn’t exist and we devoted our lives to believing he did, then nothing much was lost. But if God did exist and we had concluded He didn’t, then we were doomed for eternity. Much better to play it safe.

So, are you in a hyper-realistic simulation designed to test your loyalty to a future AI? Are you willing to risk your life by acting with the certainty you are not?

Or is it better to get a job in programming and throw yourself into the task of creating this future AI, just in case? After all, you never know whether Roko’s Basilisk is pulling the strings of your existence, or what it might do if you disappoint it.

Top Image: Roko’s Basilisk relies on the potential development of VR indistinguishable from reality, something entirely achievable from a technology standpoint (cottonbro studio / Public Domain)